hide
Free keywords:
SOIL-MOISTURE MEMORY; CLIMATE; SIMULATIONS; CIRCULATION; PERFORMANCE;
HYDROLOGY; FORECASTS; EXTREMES; SCHEMES; IMPACTEnvironmental Sciences & Ecology; Meteorology & Atmospheric Sciences; hydrological dataset; simple water balance model; soil moisture; runoff;
evapotranspiration; meteorological forcing versus parameter uncertainty;
Abstract:
Land surface hydrology can play a crucial role during extreme events such as droughts, floods and even heat waves. We introduce in this study a new hydrological dataset for Europe that consists of soil moisture, runoff and evapotranspiration (ET). It is derived with a simple water balance model (SWBM) forced with precipitation, temperature and net radiation. The SWBM dataset extends over the period 1984-2013 with a daily time step and 0.5 degrees x 0.5 degrees resolution. We employ a novel calibration approach, in which we consider 300 random parameter sets chosen from an observation-based range. Using several independent validation datasets representing soil moisture ( or terrestrial water content), ET and streamflow, we identify the best performing parameter set and hence the new dataset. To illustrate its usefulness, the SWBM dataset is compared against several state-of-the-art datasets (ERA-Interim/Land, MERRA-Land, GLDAS-2-Noah, simulations of the Community Land Model Version 4), using all validation datasets as reference. For soil moisture dynamics it outperforms the benchmarks. Therefore the SWBM soil moisture dataset constitutes a reasonable alternative to sparse measurements, little validated model results, or proxy data such as precipitation indices. Also in terms of runoff the SWBM dataset performs well, whereas the evaluation of the SWBMET dataset is overall satisfactory, but the dynamics are less well captured for this variable. This highlights the limitations of the dataset, as it is based on a simple model that uses uniform parameter values. Hence some processes impacting ET dynamics may not be captured, and quality issues may occur in regions with complex terrain. Even though the SWBM is well calibrated, it cannot replace more sophisticated models; but as their calibration is a complex task the present dataset may serve as a benchmark in future. In addition we investigate the sources of skill of the SWBM dataset and find that the parameter set has a similar impact on the simple model results as the choice of the forcing dataset. The newly derived SWBM dataset is of relevance for a range of applications given the deficit of available land datasets.
